/*
* linux/fs/fcntl.c
*
* Copyright (C) 1991, 1992 Linus Torvalds
*/
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/file.h>
#include <linux/dnotify.h>
#include <linux/smp_lock.h>
#include <linux/slab.h>
#include <asm/poll.h>
#include <asm/siginfo.h>
#include <asm/uaccess.h>
extern int sock_fcntl (struct file *, unsigned int cmd, unsigned long arg);
extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
extern int fcntl_getlease(struct file *filp);
/* Expand files. Return <0 on error; 0 nothing done; 1 files expanded,
* we may have blocked.
*
* Should be called with the files->file_lock spinlock held for write.
*/
27 static int expand_files(struct files_struct *files, int nr)
{
int err, expand = 0;
#ifdef FDSET_DEBUG
printk (KERN_ERR __FUNCTION__ " %d: nr = %d\n", current->pid, nr);
#endif
34 if (nr >= files->max_fdset) {
expand = 1;
36 if ((err = expand_fdset(files, nr)))
37 goto out;
}
39 if (nr >= files->max_fds) {
expand = 1;
41 if ((err = expand_fd_array(files, nr)))
42 goto out;
}
err = expand;
out:
#ifdef FDSET_DEBUG
if (err)
printk (KERN_ERR __FUNCTION__ " %d: return %d\n", current->pid, err);
#endif
50 return err;
}
/*
* locate_fd finds a free file descriptor in the open_fds fdset,
* expanding the fd arrays if necessary. The files write lock will be
* held on exit to ensure that the fd can be entered atomically.
*/
59 static int locate_fd(struct files_struct *files,
struct file *file, int orig_start)
{
unsigned int newfd;
int error;
int start;
write_lock(&files->file_lock);
repeat:
/*
* Someone might have closed fd's in the range
* orig_start..files->next_fd
*/
start = orig_start;
74 if (start < files->next_fd)
start = files->next_fd;
newfd = start;
78 if (start < files->max_fdset) {
newfd = find_next_zero_bit(files->open_fds->fds_bits,
files->max_fdset, start);
}
error = -EMFILE;
84 if (newfd >= current->rlim[RLIMIT_NOFILE].rlim_cur)
85 goto out;
error = expand_files(files, newfd);
88 if (error < 0)
89 goto out;
/*
* If we needed to expand the fs array we
* might have blocked - try again.
*/
95 if (error)
96 goto repeat;
98 if (start <= files->next_fd)
files->next_fd = newfd + 1;
error = newfd;
out:
104 return error;
}
107 static inline void allocate_fd(struct files_struct *files,
struct file *file, int fd)
{
FD_SET(fd, files->open_fds);
FD_CLR(fd, files->close_on_exec);
112 write_unlock(&files->file_lock);
fd_install(fd, file);
}
116 static int dupfd(struct file *file, int start)
{
struct files_struct * files = current->files;
int ret;
ret = locate_fd(files, file, start);
122 if (ret < 0)
123 goto out_putf;
allocate_fd(files, file, ret);
125 return ret;
out_putf:
128 write_unlock(&files->file_lock);
fput(file);
130 return ret;
}
133 asmlinkage long sys_dup2(unsigned int oldfd, unsigned int newfd)
{
int err = -EBADF;
struct file * file, *tofree;
struct files_struct * files = current->files;
write_lock(&files->file_lock);
140 if (!(file = fcheck(oldfd)))
141 goto out_unlock;
err = newfd;
143 if (newfd == oldfd)
144 goto out_unlock;
err = -EBADF;
146 if (newfd >= current->rlim[RLIMIT_NOFILE].rlim_cur)
147 goto out_unlock;
get_file(file); /* We are now finished with oldfd */
err = expand_files(files, newfd);
151 if (err < 0)
152 goto out_fput;
/* To avoid races with open() and dup(), we will mark the fd as
* in-use in the open-file bitmap throughout the entire dup2()
* process. This is quite safe: do_close() uses the fd array
* entry, not the bitmap, to decide what work needs to be
* done. --sct */
/* Doesn't work. open() might be there first. --AV */
/* Yes. It's a race. In user space. Nothing sane to do */
err = -EBUSY;
tofree = files->fd[newfd];
164 if (!tofree && FD_ISSET(newfd, files->open_fds))
165 goto out_fput;
files->fd[newfd] = file;
FD_SET(newfd, files->open_fds);
FD_CLR(newfd, files->close_on_exec);
170 write_unlock(&files->file_lock);
172 if (tofree)
filp_close(tofree, files);
err = newfd;
out:
176 return err;
out_unlock:
178 write_unlock(&files->file_lock);
179 goto out;
out_fput:
182 write_unlock(&files->file_lock);
fput(file);
184 goto out;
}
187 asmlinkage long sys_dup(unsigned int fildes)
{
int ret = -EBADF;
struct file * file = fget(fildes);
192 if (file)
ret = dupfd(file, 0);
194 return ret;
}
#define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | FASYNC)
199 static int setfl(int fd, struct file * filp, unsigned long arg)
{
struct inode * inode = filp->f_dentry->d_inode;
int error;
/*
* In the case of an append-only file, O_APPEND
* cannot be cleared
*/
208 if (!(arg & O_APPEND) && IS_APPEND(inode))
209 return -EPERM;
/* Did FASYNC state change? */
212 if ((arg ^ filp->f_flags) & FASYNC) {
213 if (filp->f_op && filp->f_op->fasync) {
error = filp->f_op->fasync(fd, filp, (arg & FASYNC) != 0);
215 if (error < 0)
216 return error;
}
}
/* required for strict SunOS emulation */
221 if (O_NONBLOCK != O_NDELAY)
222 if (arg & O_NDELAY)
arg |= O_NONBLOCK;
filp->f_flags = (arg & SETFL_MASK) | (filp->f_flags & ~SETFL_MASK);
226 return 0;
}
229 static long do_fcntl(unsigned int fd, unsigned int cmd,
unsigned long arg, struct file * filp)
{
long err = -EINVAL;
234 switch (cmd) {
235 case F_DUPFD:
236 if (arg < NR_OPEN) {
get_file(filp);
err = dupfd(filp, arg);
}
240 break;
241 case F_GETFD:
err = get_close_on_exec(fd);
243 break;
244 case F_SETFD:
err = 0;
set_close_on_exec(fd, arg&1);
247 break;
248 case F_GETFL:
err = filp->f_flags;
250 break;
251 case F_SETFL:
252 lock_kernel();
err = setfl(fd, filp, arg);
254 unlock_kernel();
255 break;
256 case F_GETLK:
err = fcntl_getlk(fd, (struct flock *) arg);
258 break;
259 case F_SETLK:
260 case F_SETLKW:
err = fcntl_setlk(fd, cmd, (struct flock *) arg);
262 break;
263 case F_GETOWN:
/*
* XXX If f_owner is a process group, the
* negative return value will get converted
* into an error. Oops. If we keep the
* current syscall conventions, the only way
* to fix this will be in libc.
*/
err = filp->f_owner.pid;
272 break;
273 case F_SETOWN:
274 lock_kernel();
filp->f_owner.pid = arg;
filp->f_owner.uid = current->uid;
filp->f_owner.euid = current->euid;
err = 0;
279 if (S_ISSOCK (filp->f_dentry->d_inode->i_mode))
err = sock_fcntl (filp, F_SETOWN, arg);
281 unlock_kernel();
282 break;
283 case F_GETSIG:
err = filp->f_owner.signum;
285 break;
286 case F_SETSIG:
/* arg == 0 restores default behaviour. */
288 if (arg < 0 || arg > _NSIG) {
289 break;
}
err = 0;
filp->f_owner.signum = arg;
293 break;
294 case F_GETLEASE:
err = fcntl_getlease(filp);
296 break;
297 case F_SETLEASE:
err = fcntl_setlease(fd, filp, arg);
299 break;
300 case F_NOTIFY:
err = fcntl_dirnotify(fd, filp, arg);
302 break;
303 default:
/* sockets need a few special fcntls. */
err = -EINVAL;
306 if (S_ISSOCK (filp->f_dentry->d_inode->i_mode))
err = sock_fcntl (filp, cmd, arg);
308 break;
}
311 return err;
}
314 asmlinkage long sys_fcntl(unsigned int fd, unsigned int cmd, unsigned long arg)
{
struct file * filp;
long err = -EBADF;
filp = fget(fd);
320 if (!filp)
321 goto out;
err = do_fcntl(fd, cmd, arg, filp);
fput(filp);
out:
327 return err;
}
#if BITS_PER_LONG == 32
331 asmlinkage long sys_fcntl64(unsigned int fd, unsigned int cmd, unsigned long arg)
{
struct file * filp;
long err;
err = -EBADF;
filp = fget(fd);
338 if (!filp)
339 goto out;
341 lock_kernel();
342 switch (cmd) {
343 case F_GETLK64:
err = fcntl_getlk64(fd, (struct flock64 *) arg);
345 break;
346 case F_SETLK64:
err = fcntl_setlk64(fd, cmd, (struct flock64 *) arg);
348 break;
349 case F_SETLKW64:
err = fcntl_setlk64(fd, cmd, (struct flock64 *) arg);
351 break;
352 default:
err = do_fcntl(fd, cmd, arg, filp);
354 break;
}
356 unlock_kernel();
fput(filp);
out:
359 return err;
}
#endif
/* Table to convert sigio signal codes into poll band bitmaps */
static long band_table[NSIGPOLL] = {
POLLIN | POLLRDNORM, /* POLL_IN */
POLLOUT | POLLWRNORM | POLLWRBAND, /* POLL_OUT */
POLLIN | POLLRDNORM | POLLMSG, /* POLL_MSG */
POLLERR, /* POLL_ERR */
POLLPRI | POLLRDBAND, /* POLL_PRI */
POLLHUP | POLLERR /* POLL_HUP */
};
374 static void send_sigio_to_task(struct task_struct *p,
struct fown_struct *fown,
int fd,
int reason)
{
if ((fown->euid != 0) &&
(fown->euid ^ p->suid) && (fown->euid ^ p->uid) &&
381 (fown->uid ^ p->suid) && (fown->uid ^ p->uid))
382 return;
383 switch (fown->signum) {
siginfo_t si;
385 default:
/* Queue a rt signal with the appropriate fd as its
value. We use SI_SIGIO as the source, not
SI_KERNEL, since kernel signals always get
delivered even if we can't queue. Failure to
queue in this case _should_ be reported; we fall
back to SIGIO in that case. --sct */
si.si_signo = fown->signum;
si.si_errno = 0;
si.si_code = reason & ~__SI_MASK;
/* Make sure we are called with one of the POLL_*
reasons, otherwise we could leak kernel stack into
userspace. */
398 if ((reason & __SI_MASK) != __SI_POLL)
399 BUG();
400 if (reason - POLL_IN >= NSIGPOLL)
si.si_band = ~0L;
402 else
si.si_band = band_table[reason - POLL_IN];
si.si_fd = fd;
405 if (!send_sig_info(fown->signum, &si, p))
406 break;
/* fall-through: fall back on the old plain SIGIO signal */
408 case 0:
send_sig(SIGIO, p, 1);
}
}
413 void send_sigio(struct fown_struct *fown, int fd, int band)
{
struct task_struct * p;
int pid = fown->pid;
read_lock(&tasklist_lock);
419 if ( (pid > 0) && (p = find_task_by_pid(pid)) ) {
send_sigio_to_task(p, fown, fd, band);
421 goto out;
}
423 for_each_task(p) {
int match = p->pid;
425 if (pid < 0)
match = -p->pgrp;
427 if (pid != match)
428 continue;
send_sigio_to_task(p, fown, fd, band);
}
out:
432 read_unlock(&tasklist_lock);
}
static rwlock_t fasync_lock = RW_LOCK_UNLOCKED;
static kmem_cache_t *fasync_cache;
/*
* fasync_helper() is used by some character device drivers (mainly mice)
* to set up the fasync queue. It returns negative on error, 0 if it did
* no changes and positive if it added/deleted the entry.
*/
443 int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fapp)
{
struct fasync_struct *fa, **fp;
struct fasync_struct *new = NULL;
int result = 0;
449 if (on) {
new = kmem_cache_alloc(fasync_cache, SLAB_KERNEL);
451 if (!new)
452 return -ENOMEM;
}
454 write_lock_irq(&fasync_lock);
455 for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
456 if (fa->fa_file == filp) {
457 if(on) {
fa->fa_fd = fd;
kmem_cache_free(fasync_cache, new);
460 } else {
*fp = fa->fa_next;
kmem_cache_free(fasync_cache, fa);
result = 1;
}
465 goto out;
}
}
469 if (on) {
new->magic = FASYNC_MAGIC;
new->fa_file = filp;
new->fa_fd = fd;
new->fa_next = *fapp;
*fapp = new;
result = 1;
}
out:
478 write_unlock_irq(&fasync_lock);
479 return result;
}
482 void __kill_fasync(struct fasync_struct *fa, int sig, int band)
{
484 while (fa) {
struct fown_struct * fown;
486 if (fa->magic != FASYNC_MAGIC) {
printk(KERN_ERR "kill_fasync: bad magic number in "
"fasync_struct!\n");
489 return;
}
fown = &fa->fa_file->f_owner;
/* Don't send SIGURG to processes which have not set a
queued signum: SIGURG has its own default signalling
mechanism. */
495 if (fown->pid && !(sig == SIGURG && fown->signum == 0))
send_sigio(fown, fa->fa_fd, band);
fa = fa->fa_next;
}
}
501 void kill_fasync(struct fasync_struct **fp, int sig, int band)
{
read_lock(&fasync_lock);
__kill_fasync(*fp, sig, band);
505 read_unlock(&fasync_lock);
}
508 static int __init fasync_init(void)
{
fasync_cache = kmem_cache_create("fasync cache",
sizeof(struct fasync_struct), 0, 0, NULL, NULL);
512 if (!fasync_cache)
panic("cannot create fasync slab cache");
514 return 0;
}
module_init(fasync_init)